Fruit Tree Identification Based on Multi-source Remote Sensing Image Data—Taking Pomegranate Tree as an Example

Abstract

In recent years, the development of remote sensing technology systems with medium-high spatial, temporal and spectral resolution is very fast. Remote sensing technology has been widely used in the extraction of feature elements and has achieved good results. If we could identify tree species accurately use remote sensing technology, it will be convenient and efficient to obtain the area of single fruit planting in the investigation of large-scale and multi-variety agricultural products. Then providing guidance for the planning of fruit tree and avoiding expanding planting area blindly according to price trends to bring planting economic benefits. At present, the tree species structure in the fruit growing area is complex, and it is difficult to accurately distinguish a single tree species. The existing extraction method to pomegranate cultivation area is poor effective. Therefore, based on the multi-source remote sensing image and field data, we should preprocess the remote sensing image firstly, then we do the band difference, ratio analysis, spectral index calculation spectral index change tracking, image composite and identification methods co-processing. In the last step we analyse the differences of NDVI values and spectral characteristics of remote sensing images of different tree species in different periods, then we can recognize and extract the pomegranate in fruit tree planting area. The results of the experiment indicate that the method can effectively extract the distribution information of pomegranate trees in complex areas where planting information is available. Thus it can be extended to other crop identification research, which can provide a good reference for the development of precision agriculture and intelligent agriculture management.